RESEARCH ARTICLE Using genome wide association studies to identify common QTL regions in three different genetic backgrounds based on Iberian pig breed AÂ ngel M. MartõÂnez-Montes1, Almudena FernaÂndez1, MarõÂa Muñoz1,2, Jose Luis Noguera3, Josep M. Folch4,5, Ana I. FernaÂndez1¤* a1111111111 1 Departamento de GeneÂtica Animal, Instituto Nacional de InvestigacioÂn y TecnologÂõa Agraria y Alimentaria (INIA), Madrid, Spain, 2 Centro de I+D en Cerdo IbeÂrico, Zafra, Badajoz, Spain, 3 Departament de Gen tica i a1111111111 è Millora Animal, Institut de Recerca i Tecnologia Agroalimentàries (IRTA), Lleida, Spain, 4 Departament de a1111111111 Ciència Animal i dels Aliments, Facultat de Veterinària, Universitat Autònoma de Barcelona (UAB), Bellaterra, a1111111111 Spain, 5 Plant and Animal Genomics, Centre de Recerca en Agrigenòmica (CRAG), Consorci CSIC-IRTA- a1111111111 UAB-UB, Campus UAB, Bellaterra, Spain ¤ Current address: Department of Cardiology, Hospital General Universitario Gregorio MarañoÂn, Translational research, Madrid, Spain * [email protected] OPEN ACCESS Citation: MartõÂnez-Montes AÂM, FernaÂndez A, Abstract Muñoz M, Noguera JL, Folch JM, FernaÂndez AI (2018) Using genome wide association studies to One of the major limitation for the application of QTL results in pig breeding and QTN identifica- identify common QTL regions in three different genetic backgrounds based on Iberian pig breed. tion has been the limited number of QTL effects validated in different animal material. The aim PLoS ONE 13(3): e0190184. https://doi.org/ of the current work was to validate QTL regions through joint and specific genome wide associa- 10.1371/journal.pone.0190184 tion and haplotype analyses for growth, fatness and premier cut weights in three different gen- Editor: Roberta Davoli, Universita degli Studi di etic backgrounds, backcrosses based on Iberian pigs, which has a major role in the analysis Bologna, ITALY due to its high productive relevance. The results revealed nine common QTL regions, three seg- Received: May 8, 2017 regating in all three backcrosses on SSC1, 0±3 Mb, for body weight, on SSC2, 3±9 Mb, for loin Accepted: February 26, 2018 bone-in weight, and on SSC7, 3 Mb, for shoulder weight, and six segregating in two of the three backcrosses, on SSC2, SSC4, SSC6 and SSC10 for backfat thickness, shoulder and ham Published: March 9, 2018 weights. Besides, 18 QTL regions were specifically identified in one of the three backcrosses, Copyright: © 2018 MartõÂnez-Montes et al. This is five identified only in BC_LD, seven in BC_DU and six in BC_PI. Beyond identifying and validat- an open access article distributed under the terms of the Creative Commons Attribution License, ing QTL, candidate genes and gene variants within the most interesting regions have been which permits unrestricted use, distribution, and explored using functional annotation, gene expression data and SNP identification from RNA- reproduction in any medium, provided the original Seq data. The results allowed us to propose a promising list of candidate mutations, those iden- author and source are credited. tified in PDE10A, DHCR7, MFN2 and CCNY genes located within the common QTL regions Data Availability Statement: All relevant data are and those identified near ssc-mir-103-1 considered PANK3 regulators to be further analysed. available from the European Variation Archive with the following accession number: PRJEB23068. Funding: This work was funded by Ministerio de EconomõÂa y Competitividad (MINECO) project AGL2014-56369-C2. AÂngel MartõÂnez-Montes was funded by a PhD grant from the Spanish Ministerio Introduction de Ciencia e InnovacioÂn (BES-2012-056563). The animal material was generated under the CPE03- QTL identification is one of the most relevant approaches used in livestock genomic studies in 010-C3 INIA grant. order to understand the genetic architecture that regulates complex productive traits. To date, PLOS ONE | https://doi.org/10.1371/journal.pone.0190184 March 9, 2018 1 / 21 Porcine common QTL in three different genetic backgrounds Competing interests: The authors have declared different porcine breed schemes had been used for QTL scanning, from simple designs includ- that no competing interests exist. ing purebred populations such as Pietrain, Landrace or Duroc [1,2], to more complex schemes, mating different breeds in order to compare animals with diverse phenotypes, as Duroc x Pie- train [3,4], Iberian x Landrace [5] or three-way crosses such as Duroc x (Landrace x Large White) [6]. These studies have reported a large number of QTL for different productive traits such as growth (1,328), fat composition (1,311), drip loss (1,071), average daily gain (568), average backfat thickness (332) or intramuscular fat content (244) (PigQTLdb) [7]. In spite of the great amount of QTL identified in multiple pig breeds, the application of the results in pig breeding and the identification of causal genes and mutations (QTN) has not been very successful. One of the major limitations had been the low number of available mark- ers [8,9]. However, this issue has been settled with the development of high-density genotyping platforms, which provide a high number of markers along the genome, allowing us to QTL fine-map and to conduct genome-wide analysis (GWAS) [10,11]. Another major limitation for QTN identification has been the limited number of animals employed in the analyses [12±14], reporting unreliable results that cannot be validated in different genetic backgrounds. So far, few porcine QTL regions related to productive traits have been confirmed in different animal material, some exceptions are the QTL around LEPR region for growth, fatness and meat quality traits identified in Iberian x Landrace cross [15], and validated in Iberian x Meishan cross [16] and Duroc populations [17,18], the FAT1 QTL located on SSC4 associated with fatty acid metabolism validated in Meishan x Large White, Iberian x Landrace and Wild Boar intercrosses [19±23], the QTL around MC4R for performance traits [24] and the QTL located on SSC12 for fatty acid composition was validated in different Iberian and Landrace cross populations and in purebred Duroc [17,25±28]. The aim of the current work was to validate QTL regions through GWAS analyses for growth, fatness and premier cut yields in three different genetic backgrounds F1 (Iberian x Landrace) x Landrace (BC_LD), F1 (Iberian x Duroc) x Duroc (BC_DU) and F1 (Iberian x Pietrain) x Pietrain (BC_PI) backcrosses. Here, Iberian background had a major role in the analysis due to the high productive relevance of this breed [29]. Beyond identifying and vali- dating QTL, candidate genes and polymorphism within the most interesting regions have been proposed and explored. Material and methods Animals Phenotypic and genotypic data used in this study belong to three different experimental back- crosses: F1 (Iberian x Landrace) x Landrace, F1 (Iberian x Duroc) x Duroc and F1 (Iberian x Pietrain) x Pietrain. The Iberian parental belong to different Iberian strains, Guadyerbas (black hairless strain), used for the Iberian x Landrace backcross, and Torbiscal (red strain), used for the Iberian x Duroc and Iberian x Pietrain backcrosses, which differ, apart from the coat colour, in productive traits such as growth ratio, backfat thickness and premium cut yields [30]. Schematic representation of the backcross generation is shown in Fig 1. All pigs were raised and fed under the standard, intensive system in Europe; males were not castrated. After a suckling period of between 23 and 28 d, piglets were allocated in pens with 12 individuals in each pen and were given ad libitum access to a pelleted diet (13.4 MJ/kg of ME, 18.3% of CP, 1.2% of lysine). When the piglets were about 75 d old, they were moved to a fattening building. They were penned in groups of 10 to 12 animals separated by sex, and during the whole test period they had ad libitum access to a cereal based commercial diet (13.4 MJ/kg of ME, 17.5% CP, 1% lysine). Pigs tested at the same time and in the same fattening building were considered as 1 contemporary group (batch). PLOS ONE | https://doi.org/10.1371/journal.pone.0190184 March 9, 2018 2 / 21 Porcine common QTL in three different genetic backgrounds Fig 1. Backcross generation scheme: Schematic representation of each of the backcrosses used (BC_LD, BC_DU and BC_PI), specifying the number of individuals in each generation. https://doi.org/10.1371/journal.pone.0190184.g001 PLOS ONE | https://doi.org/10.1371/journal.pone.0190184 March 9, 2018 3 / 21 Porcine common QTL in three different genetic backgrounds Ethics statement. All animal procedures were performed according to the Spanish Policy for Animal Protection RD1201/05, which meets the European Union Directive 86/609 about the protection of animals used in experimentation. The protocol was approved by the Com- mittee on the Ethics of Animal Experiments of the Instituto Nacional de InvestigacioÂn y Tec- nologõÂa Agraria y Alimentaria CEEA (Permit Number: 2014/026). Phenotypic data Seven traits related to growth, fatness and premium cut yields recorded for all three backcross pigs were analyzed (Table 1). These traits were: body weight at 150 days of mean age (BW150), backfat thickness measured at 75 kg of live weight (BFT75) and at slaughter (BFTS), mean weights (left and right) of premium cuts, hams (HW) shoulders (SW) and loin bone-in (LBW), and intramuscular fat content (IMF) measured in Longissimus dorsi samples at slaughter as described in FernaÂndez et al. [31]. Genotypic data Two different genotyping platforms were used, BC_LD and BC_PI backcrosses were genotyped with the platform PorcineSNP60 BeadChip (Illumina, Inc.) [32], containing 64,232 SNPs. Geno- meStudio software (Illumina, Inc.) was employed to visualize, edit, standard quality filter and extract genotyping data. Backcross BC_DU was genotyped with Axiom1 Porcine Genotyping Array (Affymetrix, Inc.) [33], containing 658,692 SNPs. Axiom™Analysis Suite 2.0 was employed to visualize, quality filter and extract genotype data.